OpenAI has partnered with AppsFlyer to bring mobile measurement capabilities to ChatGPT ads, allowing brands to track app installs, in-app purchases, and subscriptions directly from the platform. More than 40 brands, including Grubhub, are currently testing this integration, marking a significant step toward legitimizing ChatGPT as a serious advertising channel.

Why Has Mobile Measurement Been Missing From ChatGPT Ads?

Until now, ChatGPT ads lacked a critical piece of infrastructure that performance marketers rely on: the ability to measure what happens after someone clicks an ad. AppsFlyer is the industry standard for mobile attribution, aggregating data across channels so marketing teams can compare the cost of acquiring users and track whether those users actually make purchases or subscribe to services. Without this integration, brands couldn't justify moving significant advertising budgets to ChatGPT because they couldn't prove the platform delivered measurable results.

Attribution data isn't a luxury feature for performance marketing teams; it's the foundation on which budget allocation decisions are made. The three signals the integration now surfaces represent the metrics that determine profitability in mobile user acquisition campaigns:

• App Installs: The number of users who download an app after seeing a ChatGPT ad, though this metric alone doesn't indicate campaign success.
• In-App Purchases: Revenue generated directly from users acquired through ChatGPT ads, connecting ad spend to actual customer spending.
• Subscriptions: Recurring revenue from users who convert to paid subscriptions after installing an app from a ChatGPT ad placement.

By offering all three metrics from day one, OpenAI is signaling that it's targeting sophisticated advertisers who already run complex mobile campaigns, not just brands experimenting with a new format.

What Does This Mean for the Broader AI Advertising Landscape?

The scale of the initial test group is noteworthy. Over 40 brands in measurement testing represents a meaningful cohort for a feature that didn't exist before this integration. Grubhub's participation is particularly telling; it's a high-frequency consumer app where the path from install to order is a well-understood conversion funnel. This suggests the test spans categories where post-install revenue events are the primary success metric.

The breadth of the group also reflects how quickly the advertising industry has moved to evaluate ChatGPT as a channel since OpenAI opened its platform to advertisers. Industry conferences like Cannes Lions 2026 in June featured dedicated sessions exploring AI-native ad formats and measurement, with major brand partners examining how AI platforms fit into existing media plans. The AppsFlyer announcement brings that conversation from strategy to operational infrastructure.

How Should Marketing Teams Evaluate ChatGPT as an Ad Channel?

For teams already using AppsFlyer, the technical lift to begin measuring ChatGPT ad placements is minimal. The harder strategic questions are about media performance: how ChatGPT's user intent compares to search or social as a context for app discovery, and whether the install-to-purchase rates the platform generates justify its cost-per-thousand-impressions (CPM) against established channels like paid social or app store search.

The next critical signal to watch is whether AppsFlyer's integration is followed by announcements from other major mobile measurement partners. When multiple attribution platforms integrate with a new advertising channel, it typically indicates the platform has crossed the threshold from experimental feature to serious media channel. If Grubhub and the other 40-plus brands in the test see post-install conversion rates that compete with paid social or app store search, expect OpenAI to publicize benchmarks to accelerate broader adoption.

A Separate Insight: How ChatGPT Works Better With Rambling Than Perfect Prompts

While OpenAI refines ChatGPT's advertising capabilities, new research into how people interact with the chatbot reveals an unexpected finding: unstructured, conversational input often produces better results than carefully polished prompts. Users who spend 10 minutes talking through their thoughts with ChatGPT's voice mode, rather than crafting a single perfect question, report that the AI surfaces insights they wouldn't have discovered through traditional prompt engineering.

The reason is straightforward: when users edit themselves before asking ChatGPT something, they naturally remove contradictions, unfinished thoughts, and tangential ideas. Those details are often exactly what explains what someone really wants. ChatGPT's language model can detect patterns across rambling input that disappear in a tidy two-sentence prompt because the user has already edited them away.

This insight has practical implications for how people use ChatGPT for brainstorming, decision-making, and problem-solving. Carefully written prompts remain the best choice for tasks where you want something specific and structured. But when trying to untangle your own thoughts or explore a complex decision, letting ChatGPT listen to several minutes of unfiltered thinking often produces more useful guidance.

The convergence of these two developments, one focused on ChatGPT's role in advertising infrastructure and the other on how the model responds to natural conversation, reflects the platform's expanding role in both business operations and everyday decision-making.